I've got a question with regard to the LNK306.
In our application the output of the chip creates an oscillation around 800kHz which goes out of spec during Conducted Emissions testing.
The oscillation frequency does not drift / change over time, it is also not depending on the amount of power drawn from the output. The oscillation frequency is affected by changing components and/or the board layout(but the amplitude not so much).
By using the recommended filtering and board layout we can get the normal switching noise well below the CISPR limits but the 800kHz peak amplitude remains unchanged.
The attached FFT https://ac-dc.power.com/sites/default/files/forum/files/SCR97.PNG image shows the level of noise is around the same level of the normal switching noise but the PI filter on the input does net prevent his from getting fed back to the mains grid.
During startup there is no oscillation on the output, there is a very steady/clean switching signal until it reaches the required voltage.
The LISN measurement clearly shows a 966kHz spike above the switching signal.
When checking the design with a near field H and E probe the 800kHz signal is the strongest measured at L4 (see attached schematic).
We've changed the layout (groundplane) and the inductors (different types) but this only results in the frequency slightly changing to a higher/lower frequency but not an increase or decrease of this signal.
Can you give us any pointers with regard to origin of this oscillation in our implementation/the LNK306 and how to resolve this.
The output voltage of our circuit is 24 volt with an approximate load of 15~50mA depending on the state of the application.
The ripple on the 24V output is well below 200mV and does not contain the 800kHz oscillation.